Distance Penalties and Human Capital Returns: A Spatial Analysis of Chinese Cities

Distance Penalties and Human Capital Returns: A Spatial Analysis of Chinese Cities | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Distance Penalties and Human Capital Returns: A Spatial Analysis of Chinese Cities Zhengguang Tang, Dr Marie Ryan, Prof Eleanor Doyle, Dr Catherine Kavanagh This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7216194/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract China's coastal cities enjoy per capita GDP ten times higher than western interior cities despite decades of spatial integration policies. We examine three representative cities, Lanzhou, Guiyang, and Beijing, exemplifying 127 resource-dependent, 89 transitioning, and 31 advanced urban economies respectively from 1995-2015. Using coefficient of variation weighting and Vector Error Correction Models, we identify education returns of 1.23 in resource-dependent Lanzhou versus 0.61 in advanced Beijing, demonstrating 100% spatial differentials. Health investments show threshold effects only where mortality exceeds 6 per 1,000. Population effects vary from -8.33 in isolated western cities to +6.20 in accessible transitional economies. While our three-city analysis cannot capture spatial spillovers between cities, city-specific parameters reveal how location fundamentally shapes development relationships. These baseline findings provide quantitative benchmarks for evaluating contemporary spatial policies under China's Common Prosperity initiative. JEL Classification: R11, C31, O15, R23 Spatial heterogeneity Human capital returns Distance penalties Regional convergence China Spatial econometrics VECM Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction China's spatial development presents a fundamental puzzle: why do human capital investments yield such different returns across space within one country? Despite four decades of reforms and infrastructure investments, the ratio of per capita GDP between coastal and western cities exceeds 10:1. Beijing's educational expenditure per student of 27,896 CNY overshadows Lanzhou's 2,226 CNY, reflecting deep spatial inequalities. The literature provides competing explanations for persistent disparities. Krugman's (1991) new economic geography suggests agglomeration forces create self-reinforcing spatial concentration, whilst neoclassical models predict convergence through factor mobility. Recent empirical work reveals more complex patterns. Moretti (2011) documents skill-biased spatial sorting, whilst Glaeser and Lu (2018) find human capital externalities amplify with urban density. For China specifically, Au and Henderson (2006) identify substantial spatial inefficiencies from migration restrictions, whilst Combes et al. (2019) demonstrate how spatial sorting and agglomeration interact to generate heterogeneous returns. Existing literature predominantly examines provincial aggregates or large cross-city samples, missing crucial mechanisms that operate at finer spatial scales. Provincial analyses by Li and Huang (2009) and Fleisher et al. (2010) cannot capture within-province spatial variation, whilst large-N city studies sacrifice depth for breadth. The aggregation problem proves particularly acute given China's geography. We examine three cities representing distinct positions in China's spatial hierarchy. As Figure 1 illustrates, we select Lanzhou in northwest Gansu, Guiyang in southwest Guizhou, and Beijing on the eastern coast not as isolated cases but as exemplars of 127 resource-dependent, 89 transitioning, and 31 advanced urban economies respectively. This approach enables investigation of spatial mechanisms whilst maintaining relevance across China's urban system. Our three-city analysis prevents modelling spatial spillovers between cities or constructing spatial weight matrices typically applied in spatial econometrics. We address this by allowing all parameters to vary by location, revealing how spatial position shapes economic relationships rather than capturing network effects. While we cannot test for broader spatial autocorrelation patterns with N=3, our approach provides depth in understanding location-specific mechanisms that large-N studies miss. While spatial weight matrices based on contiguity or inverse distance would capture spillovers in larger samples, our approach reveals depth in location-specific mechanisms. Our empirical strategy employs two innovations suited to spatial analysis. We use coefficient of variation weighting to construct location-specific human capital indices, recognising that indicators may evolve differently across locations due to varying policy responsiveness and institutional constraints. We apply Vector Error Correction Modelling to identify both long-run spatial equilibrium relationships and short-run adjustment dynamics, addressing the complex endogeneity between human capital, migration, and spatial development. Our analysis covers 1995-2015, establishing baseline relationships before recent digital transformation and COVID-19. Post-2015 developments including remote work and e-commerce may alter the spatial patterns we identify. We explicitly frame this as a baseline study providing benchmarks for evaluating how contemporary policies affect spatial development. The findings reveal spatial heterogeneity that extends spatial economics theory. Education returns vary from 1.2 in distant Lanzhou to 0.6 in accessible Beijing, representing a 100% differential exceeding most cross-country estimates. This spatial gradient reflects not just different development levels but fundamental differences in how human capital translates to productivity across space. Health reveals spatial threshold effects, generating returns only in locations where basic mortality constraints bind. Population dynamics reveal how distance penalties interact with human capital, with brain drain intensifying with education in isolated cities whilst accessible transitioning cities attract skilled workers. 2. Theoretical Framework and Spatial Context 2.1 Spatial Foundations of Human Capital Returns Human capital theory traditionally abstracts from spatial considerations yet returns fundamentally depend on spatial context through multiple channels. Our framework integrates human capital with spatial economics to examine heterogeneous returns across China's urban hierarchy. Marshall's (1890) agglomeration economies gain new relevance when combined with human capital theory. In spatial equilibrium, educated workers sort into dense locations where knowledge spillovers maximise returns (Glaeser and Resseger, 2010). The productivity of any individual depends not just on their own human capital but on the human capital of those around them: Equation (1): yi = Ac hi^α Hc^β where yi is individual productivity, Ac is city-specific total factor productivity, hi is individual human capital, Hc is aggregate city human capital, and β captures spillovers. This implies returns to education increase with city human capital density, creating divergent spatial dynamics. The gravity structure of economic interactions imposes distance penalties on isolated locations. Following Redding and Venables (2004), market access affects productivity through: Equation (2): MAc = Σk τck^(-σ) Yk where τck represents transport costs between cities c and k, and σ is the elasticity of substitution. For human capital, distance penalties operate through limited access to ideas, technologies, and specialised inputs that complement education. This suggests returns diminish with distance from economic centres. We incorporate market access directly in our econometric specification to capture these mechanisms. Spatial equilibrium requires equalising utility across locations, but heterogeneous preferences and abilities create sorting patterns (Behrens et al., 2014). High-ability workers select into high-cost, high-productivity cities, whilst mobility costs vary by skill level. China's hukou system adds institutional friction, creating spatial misallocation that may paradoxically increase returns in constrained locations (Tombe and Zhu, 2019). 2.2 China's Spatial Development Context Western China faces severe geographic barriers including the Tibetan Plateau, Taklamakan Desert, and mountain ranges featuring natural isolation. Transport costs remain high despite infrastructure investments, with freight costs per kilometre in western regions exceeding eastern regions by 40-60% (World Bank, 2019). The hukou system segments labour markets. Approximately 290 million migrants n cities without local hukou, face exclusion from public services. This institutional friction interacts with distance, as remote cities cannot attract migrants even with high wages. Spatial development policies from Western Development launched in 2000 to Belt and Road initiated in 2013 attempted to overcome geographic disadvantages through infrastructure and subsidies. Evaluations are mixed with indications of reduced efficiency from supporting locations with fundamental disadvantages (Zhu, et al., 2019). 2.3 Representative Cities in Spatial Perspective Our three cities represent distinct positions in China's spatial hierarchy, enabling analysis of how location shapes development possibilities. Lanzhou exemplifies cities facing severe distance penalties. Located 1,600km from the coast in the narrow Hexi Corridor, market access show Lanzhou's effective economic distance exceeds geographic distance given limited transport options. Only one major railway and highway connect it to eastern markets. Its industrial structure reflects transport costs, with 37% of GDP from heavy industry including petrochemicals and non-ferrous metals with high value-to-weight ratios capable of bearing transport costs. Despite hosting several universities, Lanzhou suffers persistent brain drain as graduates migrate to accessible cities. The spatial mismatch between human capital supply from universities and limited demand from high-skill employers exemplifies challenges facing 127 similar isolated cities across China's western interior. Guiyang occupies an intermediate position, located 800km from major ports but connected to the Pearl River Delta through improved transport infrastructure. High-speed rail introduced in 2014 reduced travel time to Guangzhou from Beijing by 80% - to 4 hours. Improved accessibility coincided with economic transformation as services expanded from 42% to 56% of GDP during our study period. The city leverages its position between China's poorest rural areas and wealthy coastal cities. As Guizhou's capital, it serves as a gateway for rural-urban migrants, processing over 2 million temporary migrants annually. Its intermediary role means it both loses educated workers to coastal areas and gains them from rural hinterlands. Big data initiatives explicitly leverage Guiyang's position, offering lower costs than coastal cities whilst maintaining digital connectivity. Beijing epitomises spatial agglomeration advantages. As the political capital with premier universities and corporate headquarters, it generates self-reinforcing concentration. The city's effective market access extends nationally through political and corporate networks. With 26% of China's research personnel despite having 1.6% of population, Beijing demonstrates extreme spatial concentration of high-skill activities. Agglomeration economies manifest through thick labour markets with 750,000 university graduates annually, knowledge spillovers generating 45% of national patents in some technology classes, and specialised inputs including venture capital and professional services. Agglomeration also creates offsetting congestion costs including severe air pollution, traffic gridlock (average commutes of 52 minutes), and housing costs claiming 58% of average income. 3. Methods 3.1 Data Sources and Spatial Considerations We use panel data from 1995-2015 compiled from Statistical Yearbooks of Lanzhou, Guiyang, and Beijing. This period captures China's spatial transformation including migration flows of 150 million rural-urban migrants, transport infrastructure expansion that quadrupled the highway network, and evolving spatial policies. The timeframe provides sufficient variation to identify spatial mechanisms whilst maintaining institutional stability. All monetary variables are deflated using city-specific Consumer Price Indices, crucial given spatially heterogeneous inflation. Services in Beijing inflated 40% faster than goods in Lanzhou. Spatial price differences matter as identical nominal salaries represent vastly different real compensation across locations due to housing and service costs. 3.2 Measuring Human Capital in Spatial Context We construct education and health indices using coefficient of variation weighting to reflect local conditions. For each indicator i in city j, we calculate: Equation (3): CVij = σij / μij Where σij is the standard deviation of indicator i in city j across the time period:μij is the mean of that indicator over time. This temporal CV within each city reveals policy responsiveness given spatial constraints. An indicator stable in accessible Beijing but volatile in isolated Lanzhou suggests different development possibilities across space. The approach captures how spatial position shapes the ability to implement change. Education indicators span inputs including expenditure and salaries, processes including enrolment and infrastructure, and outputs including graduation rates, each affected differently by spatial factors. Distance increases costs through teacher recruitment difficulty whilst reducing benefits through graduate retention challenges. Health indicators similarly reflect spatial challenges as medical personnel concentrate in accessible cities whilst isolated areas struggle with basic provision. 3.3 Spatial Econometric Considerations Our VECM specification incorporates spatial elements: Equation (4): Δyjt = αj + γj(yj,t-1 - β'xj,t-1) + Σi=1^p-1 ΓjiΔyj,t-i + δjMAjt + νjt where Δyjt represents the change in variables for city j at time t, αj captures city-specific constants, γj is the error correction coefficient indicating adjustment speed to long-run equilibrium, β contains long-run relationship parameters, Γji captures short-run dynamics, MAjt represents market access from equation (2), and νjt is the error term. Market access for each city is calculated as the sum of distance-weighted GDP of all prefecture-level cities within 500km radius. All parameters (α, γ, β, Γ, δ) vary by city j, capturing how spatial position shapes economic relationships. While we cannot estimate spatial lag terms or test for spatial autocorrelation patterns with three cities, this specification allows spatial heterogeneity in all relationships. 4. Results 4.1 Spatial Patterns in Human Capital Development Figure 1 illustrates the spatial configuration of our study cities. Lanzhou is located in northwest Gansu along the narrow Hexi Corridor, Guiyang in mountainous southwest Guizhou, and Beijing on the eastern coast, highlighting differing positions relative to major transport corridors and economic zones. Figure 2 reveals the spatial gradient in education resources, fundamental to understanding differential returns. Beijing's expenditure increased from approximately 12,000 to 50,000 CNY, whilst Lanzhou and Guiyang remain below 10,000 throughout the period. The tenfold expenditure gap reflects more than fiscal capacity. It represents the spatial concentration of quality education. Beijing attracts top teachers through 70% salary premiums, hosts elite schools, and benefits from corporate training programmes. This spatial inequality in education quality amplifies raw expenditure differences. Figure 3 illustrates demographic pressures varying across space. The figure shows Lanzhou and Guiyang with 800-1,100 students per 10,000 declining after 2007, whilst Beijing maintains 300-400 throughout. The spatial gradient in youth dependency reflects differential migration patterns. Beijing attracts working-age migrants whilst their children often remain in origin cities due to hukou restrictions, creating divergent demographic pressures. Interior cities must provide education for larger youth cohorts with fewer resources, representing a fundamental spatial inequality. 4.2 Coefficient of Variation Analysis Table 1 reveals how different indicators drive change across spatial contexts. Table 1: Coefficient of Variation (CV) Results for Education Indicators Indicator Lanzhou Guiyang Beijing SD Mean CV SD Mean CV SD Mean CV Expenditure per student 1,913 2,226 0.86 2,086 2,420 0.86 10,241 27,896 0.37 Teacher salary 16,875 22,199 0.76 17,607 24,025 0.73 30,365 41,355 0.73 Primary students/10k 164 837 0.20 143 977 0.15 172 469 0.37 University students/10k 465 690 0.67 211 494 0.43 442 693 0.64 Sum of CV 9.74 9.70 10.19 Source: Authors' calculations. Full results for all 14 indicators available upon request. The CV patterns reveal spatial constraints on development. Financial indicators show high variation in interior cities as governments struggle to increase funding from low bases. Beijing's lower expenditure CV of 0.37 reflects stable high investment. Demographic indicators vary more in Beijing due to migration volatility, revealing how accessibility creates different development challenges. 4.3 Human Capital Returns Across Space Figure 4 demonstrates the remarkable convergence in education indices despite persistent resource gaps. The figure shows indices converging from initial gaps where Beijing starts at 47, Lanzhou at 29, and Guiyang at 27, to a narrow final range of 65-73. Convergence masks fundamental spatial differences in what drives improvement. Interior cities achieved gains through basic quantity expansion including more schools, teachers, and enrolment. Beijing improved through quality enhancement including better teacher training, curriculum innovation, and technology integration. The indices converge whilst the nature of education diverges spatially. 4.4 Econometric Evidence of Spatial Heterogeneity Table 2 presents our core findings on spatially differentiated returns. Table 2: Long-Run VECM Coefficients Revealing Spatial Heterogeneity Variable Lanzhou Guiyang Beijing ln(Population) -8.33*** 6.20*** 1.02*** (2.19) (2.20) (0.40) ln(Education) 1.23*** 0.93*** 0.61** (0.47) (0.31) (0.13) ln(Health) 0.27** -0.27 -0.10 (0.12) (0.22) (0.15) ln(Fixed Assets) 0.74*** 0.30*** 0.77*** (0.19) (0.07) (0.10) ln(R&D) 0.02 0.22*** 0.47*** (0.08) (0.03) (0.06) *Note: Standard errors in parentheses. **, ** denote 1%, 5% significance. Source: Authors' VECM estimations. The results reveal spatial heterogeneity. The gradient from 1.23 in isolated Lanzhou to 0.61 in accessible Beijing demonstrates how distance amplifies returns. In remote locations, educated workers are scarce, creating high marginal productivity. The 100% differential suggests spatial inefficiency where the same education investment generates double the growth impact in lagging regions. Population coefficients from -8.3 to +6.2 reveal how space shapes migration-development relationships. Lanzhou's negative coefficient indicates vicious cycles where education enables out-migration, undermining local development. Guiyang's positive coefficient suggests it has reached spatial thresholds where improved human capital attracts rather than repels talent. Beijing's modest effect reflects spatial equilibrium where in-migration benefits balance congestion costs. R&D returns increase monotonically with spatial position from 0.02 to 0.47, confirming agglomeration theories. Innovation requires dense networks of researchers, specialised suppliers, and knowledge institutions that concentrate spatially. Isolated cities cannot generate returns from R&D without the spatial preconditions. 5. Discussion 5.1 Theoretical Implications for Spatial Economics Our findings extend spatial economic theory. The 100% education return differential between remote and accessible cities quantifies distance penalties beyond transport costs. This gradient reflects how distance limits idea flows, reduces matching quality in labour markets, and constrains specialisation possibilities. Standard spatial models focusing on goods transport miss these knowledge-distance interactions crucial for understanding modern spatial inequality. Beijing's lower education returns appearparadoxical through an agglomeration lens. However, this reflects spatial equilibrium where agglomeration benefits capitalise into land prices and congestion costs, reducing net returns. The spatial sorting of high-ability workers to Beijing means marginal education investments have less impact than in human capital-scarce locations. Health returns appearing only in high-mortality locations reveal spatial development thresholds. Basic infrastructure must reach minimum levels before locations can exploit agglomeration advantages. This sequencing of first overcoming fundamental disadvantages, then building specialisation, challenges models assuming smooth spatial transitions. The interaction between education and migration in remote cities creates spatial poverty traps missed by standard models. Education investments intended to develop lagging regions may accelerate divergence by facilitating out-migration. This paradox requires reconceptualising spatial policy. Isolated investments fail without complementary improvements in spatial connectivity and local opportunity. 5.2 Policy Implications Our baseline evidence provides quantitative guidance for spatially differentiated policies. The 100% return differential strongly supports Common Prosperity's emphasis on spatial redistribution. Each CNY invested in education in cities like Lanzhou generates roughly double the return compared to Beijing. However, population results warn against isolated interventions. Successful spatial policy targets education, employment, and connectivity simultaneously. Supporting literature confirms these patterns. An (2018) finds that rural teachers earn wages 7-41% lower than the provincial average, with one study showing highest teacher wages in Shanghai at RMB 62,300 per year compared to Henan Province at RMB 12,800 per year. World Bank (2022) analysis of Chinese education spending shows coastal provinces spend 3-5 times more per student than western provinces. Our findings suggest spatially differentiated investment sequences. Remote cities like Lanzhou require priority on reducing distance penalties through e.g. digital infrastructure and transport links before major human capital investments. Transitional cities like Guiyang benefit from balanced packages combining moderate education and R&D with connectivity improvements. Agglomerated cities like Beijing can focus on managing congestion whilst maintaining innovation advantages. Post-2015 digital transformation may fundamentally alter spatial relationships. Remote work capabilities could reduce distance penalties for knowledge workers. E-commerce platforms may overcome market access constraints for isolated regions. Digital education could narrow quality gaps between locations. However, agglomeration advantages in innovation and high-skill services may intensify, creating new forms of spatial inequality. Future research should examine whether digital transformation reduces or reshapes spatial heterogeneity. 5.3 Methodological Contributions to Spatial Analysis The representative city approach offers advantages. By selecting cities representing broader spatial categories, we offer analytical depth whilst maintaining external validity. The 247 cities sharing our representatives' characteristics can apply insights from this analysis. Large-N spatial econometrics identify correlations yet struggle with mechanisms. Our representative-selection approach reveals how distance creates teacher recruitment challenges, why health thresholds matter spatially, and when migration reinforces versus undermines development. Researchers may identify 3-5 cities representing key spatial positions including remote resource regions, intermediate manufacturing, and connected services, then trace mechanisms through more detailed comparison. 6. Conclusions This paper reveals spatial heterogeneity in human capital returns across China's urban hierarchy. Examining three representative cities (representative of 247 urban economies occupying different spatial positions), we quantify how distance, density, and connectivity shape development possibilities. Education returns varying from 1.23 in remote Lanzhou to 0.61 in accessible Beijing demonstrate 100% spatial differentials that challenge uniform national policies. These patterns reflect not simply different development levels but fundamental spatial mechanisms including distance penalties limiting idea flows, agglomeration advantages creating congestion, and migration dynamics that can either reinforce or undermine local development. Table 3 synthesises our key findings, connecting them to existing literature and policy. Table 3: Summary of Findings, Literature Links, and Policy Implications Finding Our Results Supporting Literature Policy Connection Education return differentials 100% higher returns in Lanzhou (1.23) vs Beijing (0.61) Fleisher et al. (2010): 60% provincial gaps; Gennaioli et al. (2013): 0.3-1.8 range internationally. Supports targeted western investment; validates 3-5x spending gaps identified by World Bank (2022). Health threshold effects Returns only where mortality >6/1,000 (Lanzhou: 0.27) Bloom et al. (2004): basic health as growth prerequisite; Shen, Q., Chang et al., (2020) similar mortality thresholds. Healthy China 2030: Focus basic health in high-mortality areas before quality improvements. Population-migration traps -8.33 in Lanzhou, +6.20 in Guiyang, +1.02 in Beijing. Tombe & Zhu (2019): hukou creates spatial misallocation; Au & Henderson (2006): migration restrictions reduce efficiency. Hukou reform priorities: Bundle education with local opportunities to prevent brain drain. Distance-R&D interaction Zero returns in Lanzhou, 0.47 in Beijing. Moretti (2011): innovation requires density; Krugman (1991): agglomeration necessary for knowledge spillovers. Infrastructure must precede innovation investment in corridor cities. Spatial convergence patterns Education indices converge while quality diverges. Li & Huang (2009): provincial convergence; our city-level analysis reveals quality divergence. Western Development Strategy: Quantity targets achieved, shift focus to quality enhancement. Source: Authors' synthesis of empirical findings with literature and policy documents. Our representative city analysis provides spatial insights that complement large-N spatial econometric studies, demonstrating the value of mixed methodological approaches in spatial economics. While large samples identify average effects, representative analysis points to spatial mechanisms and thresholds crucial for policy design. The coefficient of variation approach captures how spatial position shapes policy responsiveness, whilst our econometric strategy allows all parameters to vary spatially. Our three-city analysis cannot model explicit spatial spillovers or test for spatial autocorrelation patterns that larger samples would permit. This limitation means our findings represent location-specific relationships rather than spatially integrated effects. Future research should employ explicit spatial econometric models with spatial weight matrices to capture network dependencies our approach cannot address. Specifically, future research should employ spatial panel models with contiguity-based weight matrices to capture inter-city spillovers and test for spatial autocorrelation in human capital returns. The findings establish quantitative benchmarks for evaluating spatial policies under Common Prosperity and Belt and Road initiatives. Substantial return differentials support spatial targeting, but population dynamics warn against isolated interventions.. The threshold nature of returns suggests spatial sequencing where regions must first overcome basic disadvantages, then build human capital, and finally develop innovation capacity. Post-2015 digital transformation including remote work, e-commerce, and online education may reshape distance penalties and agglomeration advantages. Whether these impact spatial inequality or create new forms of concentration remains an empirical question requiring updated analysis. Our methodology provides a replicable framework for analysing spatial heterogeneity in other developing countries facing regional inequality. Researchers can identify 3-5 cities representing key spatial positions, apply coefficient of variation weighting to capture local dynamics, and use VECM to identify spatially varying relationships. This approach is particularly valuable for countries with diverse geography where aggregate analysis masks crucial spatial variation. As China and other nations pursue spatially balanced development, recognising and responding to spatial heterogeneity rather than implementing spatially blind policies offers a promising path toward inclusive growth. References An, X. H. (2018). Teacher salaries and the shortage of high-quality teachers in China's rural primary and secondary schools. Chinese Education & Society, 51(2), 103-116. https://doi.org/10.1080/10611932.2018.1433411 Au, C. C., & Henderson, J. V. (2006). Are Chinese cities too small? Review of Economic Studies, 73(3), 549-576. https://doi.org/10.1111/j.1467-937X.2006.00387.x Behrens, K., Duranton, G., & Robert-Nicoud, F. (2014). Productive cities: Sorting, selection, and agglomeration. Journal of Political Economy, 122(3), 507-553. https://doi.org/10.1086/675534 Bloom, D. E., Canning, D., & Sevilla, J. (2004). The effect of health on economic growth: A production function approach. World Development, 32(1), 1-13. https://doi.org/10.1016/j.worlddev.2003.07.002 Combes, P. P., Démurger, S., & Li, S. (2019). Migration externalities in Chinese cities. European Economic Review, 76, 152-167. https://doi.org/10.1016/j.euroecorev.2015.02.004 Fleisher, B., Li, H., & Zhao, M. Q. (2010). Human capital, economic growth, and regional inequality in China. Journal of Development Economics, 92(2), 215-231. https://doi.org/10.1016/j.jdeveco.2009.01.010 Gennaioli, N., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2013). Human capital and regional development. Quarterly Journal of Economics, 128(1), 105-164. https://doi.org/10.1093/qje/qjs050 Glaeser, E. L., & Lu, M. (2018). Human capital externalities in China. NBER Working Paper No. 24925. https://doi.org/10.3386/w24925 Glaeser, E. L., & Resseger, M. G. (2010). The complementarity between cities and skills. Journal of Regional Science, 50(1), 221-244. https://doi.org/10.1111/j.1467-9787.2009.00635.x Krugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99(3), 483-499. https://doi.org/10.1086/261763 Li, H., & Huang, L. (2009). Health, education, and economic growth in China: Empirical findings and implications. China Economic Review, 20(3), 374-387. https://doi.org/10.1016/j.chieco.2008.05.001 Liu, J., Wang, M., Yang, L., Rahman, S., & Sriboonchitta, S. (2024). Multidimensional spatial inequality in China and its relationship with economic growth. Humanities and Social Sciences Communications, 11(1), 1-15. https://doi.org/10.1057/s41599-024-03961-y Marshall, A. (1890). Principles of Economics. London: Macmillan. Moretti, E. (2011). Local labor markets. In Handbook of Labor Economics (Vol. 4, pp. 1237-1313). Elsevier. https://doi.org/10.1016/S0169-7218(11)02412-9 Redding, S., & Venables, A. J. (2004). Economic geography and international inequality. Journal of International Economics, 62(1), 53-82. https://doi.org/10.1016/j.jinteco.2003.07.001 Shen, Q., Chang, B., Yin, G., & Wang W. (2020). The Impact of Health Investment on Economic Growth: Evidence from China. Iranian Journal of Public Health , 49 (4), 684–692. Tombe, T., & Zhu, X. (2019). Trade, migration, and productivity: A quantitative analysis of China. American Economic Review, 109(5), 1843-1872. https://doi.org/10.1257/aer.20150811 Wei, Y. D. (2000). Regional Development in China: States, Globalization and Inequality. Routledge. World Bank (2019). China Transport Sector Assessment. Washington, DC: World Bank. World Bank (2022). China - Gansu Technical and Vocational Education and Training Project. Washington, DC: World Bank. Zhu, J., Zhu, M., & Xiao, Y. (2019). Urbanization for rural development: Spatial paradigm shifts toward inclusive urban-rural integrated development in China. Journal of Rural Studies , 71 , 94–103. https://doi.org/10.1016/j.jrurstud.2019.08.009 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7216194","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":491834784,"identity":"a1516b14-ffd3-4680-a02a-e06a1a92466c","order_by":0,"name":"Zhengguang Tang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Zhengguang","middleName":"","lastName":"Tang","suffix":""},{"id":491834785,"identity":"8f6bb0ae-e981-4cea-b41c-61ae0f6264a9","order_by":1,"name":"Dr Marie Ryan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0UlEQVRIiWNgGAWjYBACgwMMDAcY/wBZ7A0gvgRhLZZgLW1AFs8BIrXYgxSCtUgkEOkws+NnHx742HZY3lzyjeGHHwwWcoS1nEk3ODjzz2HDnbNzjCV7GCSMCWs5kMZwmPfDYcYNt3PMmIHOS2wgpMXg/DOGw39nHLbfcPMMWEs9YS03gLYw9hxO3HCDB6wlgaDDDG48YzjY25aevOFMWrFkj4GEIREOS2P+8LPN2nbD8cMbP/yoqJMnaAsUNMNMIFYDA0Md8UpHwSgYBaNg5AEASXBDn+Mq/IYAAAAASUVORK5CYII=","orcid":"","institution":"University College Cork","correspondingAuthor":true,"prefix":"Dr","firstName":"Marie","middleName":"","lastName":"Ryan","suffix":""},{"id":491834786,"identity":"ecbd2a31-d4e4-4736-9cee-fca4db64e334","order_by":2,"name":"Prof Eleanor Doyle","email":"","orcid":"","institution":"University College Cork","correspondingAuthor":false,"prefix":"","firstName":"Prof","middleName":"Eleanor","lastName":"Doyle","suffix":""},{"id":491834787,"identity":"1f1582bd-91ab-4e9b-b28d-92bcce705506","order_by":3,"name":"Dr Catherine Kavanagh","email":"","orcid":"","institution":"University College Cork","correspondingAuthor":false,"prefix":"Dr","firstName":"Catherine","middleName":"","lastName":"Kavanagh","suffix":""}],"badges":[],"createdAt":"2025-07-25 16:53:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7216194/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7216194/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98204550,"identity":"27aa808d-0d6b-4fc1-88ce-4a764c840041","added_by":"auto","created_at":"2025-12-15 08:28:19","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":240173,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGeographic Location of Study Cities in China's Spatial Economy\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSource: Authors' compilation. Cities shown with their respective economic zones.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7216194/v1/1c3f9dfd9cd4cb51de5df97a.png"},{"id":98432722,"identity":"68724a3d-4eec-4633-b1e2-c24d5c11a607","added_by":"auto","created_at":"2025-12-17 16:49:50","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":17492,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEducational Expenditure per Student in Lanzhou, Guiyang, and Beijing, 1995-2015\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSource: Authors' calculations from Statistical Yearbooks of Lanzhou, Guiyang, and Beijing.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7216194/v1/17d5d2a4f7e7d6617347cbac.png"},{"id":98204552,"identity":"12508355-c8c0-4cc5-bba3-c1f0a57af583","added_by":"auto","created_at":"2025-12-15 08:28:19","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":20416,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePrimary School Students per 10,000 Population in Lanzhou, Guiyang, and Beijing, 1995-2015\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSource: Authors' calculations from Statistical Yearbooks of Lanzhou, Guiyang, and Beijing.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7216194/v1/412d539ebee3640436598d80.png"},{"id":98431185,"identity":"e5681f92-ea02-4ef4-8f5d-f4e307fb6260","added_by":"auto","created_at":"2025-12-17 16:47:14","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":24322,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEducation Index Evolution in Lanzhou, Guiyang, and Beijing, 1995-2015\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSource: Authors' calculations using CV-weighted indicators.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7216194/v1/b4da708ce4f38372fb325e67.png"},{"id":98444900,"identity":"f71d554a-ff6d-498e-bdc4-5a90932a442b","added_by":"auto","created_at":"2025-12-17 17:18:10","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1218763,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7216194/v1/2743e766-e631-42fc-8fbe-cdef55914fa6.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Distance Penalties and Human Capital Returns: A Spatial Analysis of Chinese Cities","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eChina's spatial development presents a fundamental puzzle: why do human capital investments yield such different returns across space within one country? Despite four decades of reforms and infrastructure investments, the ratio of per capita GDP between coastal and western cities exceeds 10:1. Beijing's educational expenditure per student of 27,896 CNY overshadows Lanzhou's 2,226 CNY, reflecting deep spatial inequalities.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe literature provides competing explanations for persistent disparities. Krugman's (1991) new economic geography suggests agglomeration forces create self-reinforcing spatial concentration, whilst neoclassical models predict convergence through factor mobility. Recent empirical work reveals more complex patterns. Moretti (2011) documents skill-biased spatial sorting, whilst Glaeser and Lu (2018) find human capital externalities amplify with urban density. For China specifically, Au and Henderson (2006) identify substantial spatial inefficiencies from migration restrictions, whilst Combes et al. (2019) demonstrate how spatial sorting and agglomeration interact to generate heterogeneous returns.\u003c/p\u003e\n\u003cp\u003eExisting literature predominantly examines provincial aggregates or large cross-city samples, missing crucial mechanisms that operate at finer spatial scales. Provincial analyses by Li and Huang (2009) and Fleisher et al. (2010) cannot capture within-province spatial variation, whilst large-N city studies sacrifice depth for breadth. The aggregation problem proves particularly acute given China's geography.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe examine three cities representing distinct positions in China's spatial hierarchy. As Figure 1 illustrates, we select Lanzhou in northwest Gansu, Guiyang in southwest Guizhou, and Beijing on the eastern coast not as isolated cases but as exemplars of 127 resource-dependent, 89 transitioning, and 31 advanced urban economies respectively. This approach enables investigation of spatial mechanisms whilst maintaining relevance across China's urban system. Our three-city analysis prevents modelling spatial spillovers between cities or constructing spatial weight matrices typically applied in spatial econometrics. We address this by allowing all parameters to vary by location, revealing how spatial position shapes economic relationships rather than capturing network effects. While we cannot test for broader spatial autocorrelation patterns with N=3, our approach provides depth in understanding location-specific mechanisms that large-N studies miss. While spatial weight matrices based on contiguity or inverse distance would capture spillovers in larger samples, our approach reveals depth in location-specific mechanisms.\u003c/p\u003e\n\u003cp\u003eOur empirical strategy employs two innovations suited to spatial analysis. We use\u0026nbsp;coefficient of variation weighting to construct location-specific human capital indices, recognising that indicators may evolve differently across locations due to varying policy responsiveness and institutional constraints. We apply Vector Error Correction Modelling to identify both long-run spatial equilibrium relationships and short-run adjustment dynamics, addressing the complex endogeneity between human capital, migration, and spatial development. Our analysis covers 1995-2015, establishing baseline relationships before recent digital transformation and COVID-19. Post-2015 developments including remote work and e-commerce may alter the spatial patterns we identify. We explicitly frame this as a baseline study providing benchmarks for evaluating how contemporary policies affect spatial development.\u003c/p\u003e\n\u003cp\u003eThe findings reveal spatial heterogeneity that extends spatial economics theory. Education returns vary from 1.2 in distant Lanzhou to 0.6 in accessible Beijing, representing a 100% differential exceeding most cross-country estimates. This spatial gradient reflects not just different development levels but fundamental differences in how human capital translates to productivity across space. Health reveals spatial threshold effects, generating returns only in locations where basic mortality constraints bind. Population dynamics reveal how distance penalties interact with human capital, with brain drain intensifying with education in isolated cities whilst accessible transitioning cities attract skilled workers.\u003c/p\u003e"},{"header":"2. Theoretical Framework and Spatial Context","content":"\u003cp\u003e\u003cstrong\u003e2.1 Spatial Foundations of Human Capital Returns\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHuman capital theory traditionally abstracts from spatial considerations yet returns fundamentally depend on spatial context through multiple channels. Our framework integrates human capital with spatial economics to examine heterogeneous returns across China\u0026apos;s urban hierarchy. Marshall\u0026apos;s (1890) agglomeration economies gain new relevance when combined with human capital theory. In spatial equilibrium, educated workers sort into dense locations where knowledge spillovers maximise returns (Glaeser and Resseger, 2010). The productivity of any individual depends not just on their own human capital but on the human capital of those around them:\u003c/p\u003e\n\u003cp\u003eEquation (1): yi = Ac hi^\u0026alpha; Hc^\u0026beta;\u003c/p\u003e\n\u003cp\u003ewhere yi is individual productivity, Ac is city-specific total factor productivity, hi is individual human capital, Hc is aggregate city human capital, and \u0026beta; captures spillovers. This implies returns to education increase with city human capital density, creating divergent spatial dynamics.\u003c/p\u003e\n\u003cp\u003eThe gravity structure of economic interactions imposes distance penalties on isolated locations. Following Redding and Venables (2004), market access affects productivity through:\u003c/p\u003e\n\u003cp\u003eEquation (2): MAc = \u0026Sigma;k \u0026tau;ck^(-\u0026sigma;) Yk\u003c/p\u003e\n\u003cp\u003ewhere \u0026tau;ck represents transport costs between cities c and k, and \u0026sigma; is the elasticity of substitution. For human capital, distance penalties operate through limited access to ideas, technologies, and specialised inputs that complement education. This suggests returns diminish with distance from economic centres. We incorporate market access directly in our econometric specification to capture these mechanisms.\u003c/p\u003e\n\u003cp\u003eSpatial equilibrium requires equalising utility across locations, but heterogeneous preferences and abilities create sorting patterns (Behrens et al., 2014). High-ability workers select into high-cost, high-productivity cities, whilst mobility costs vary by skill level. China\u0026apos;s hukou system adds institutional friction, creating spatial misallocation that may paradoxically increase returns in constrained locations (Tombe and Zhu, 2019).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 China\u0026apos;s Spatial Development Context\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWestern China faces severe geographic barriers including the Tibetan Plateau, Taklamakan Desert, and mountain ranges featuring natural isolation. Transport costs remain high despite infrastructure investments, with freight costs per kilometre in western regions exceeding eastern regions by 40-60% (World Bank, 2019).\u003c/p\u003e\n\u003cp\u003eThe hukou system segments labour markets. Approximately 290 million migrants n cities without local hukou, face exclusion from public services. This institutional friction interacts with distance, as remote cities cannot attract migrants even with high wages. Spatial development policies from Western Development launched in 2000 to Belt and Road initiated in 2013 attempted to overcome geographic disadvantages through infrastructure and subsidies. Evaluations are mixed \u0026nbsp;with indications of \u0026nbsp;reduced efficiency from supporting locations with fundamental disadvantages (Zhu, et al., 2019).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Representative Cities in Spatial Perspective\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur three cities represent distinct positions in China\u0026apos;s spatial hierarchy, enabling analysis of how location shapes development possibilities.\u003c/p\u003e\n\u003cp\u003eLanzhou exemplifies cities facing severe distance penalties. Located 1,600km from the coast in the narrow Hexi Corridor, market access show Lanzhou\u0026apos;s effective economic distance exceeds geographic distance given limited transport options. Only one major railway and highway connect it to eastern markets. Its industrial structure reflects transport costs, with 37% of GDP from heavy industry including petrochemicals and non-ferrous metals with high value-to-weight ratios capable of bearing transport costs. Despite hosting several universities, Lanzhou suffers persistent brain drain as graduates migrate to accessible cities. The spatial mismatch between human capital supply from universities and limited demand from high-skill employers exemplifies challenges facing 127 similar isolated cities across China\u0026apos;s western interior.\u003c/p\u003e\n\u003cp\u003eGuiyang occupies an intermediate position, located 800km from major ports but connected to the Pearl River Delta through improved transport infrastructure. High-speed rail introduced in 2014 reduced travel time to Guangzhou from Beijing by 80% - to 4 hours. Improved accessibility coincided with economic transformation as services expanded from 42% to 56% of GDP during our study period. The city leverages its position between China\u0026apos;s poorest rural areas and wealthy coastal cities. As Guizhou\u0026apos;s capital, it serves as a gateway for rural-urban migrants, processing over 2 million temporary migrants annually. Its intermediary role means it both loses educated workers to coastal areas and gains them from rural hinterlands. Big data initiatives explicitly leverage Guiyang\u0026apos;s position, offering lower costs than coastal cities whilst maintaining digital connectivity.\u003c/p\u003e\n\u003cp\u003eBeijing epitomises spatial agglomeration advantages. As the political capital with premier universities and corporate headquarters, it generates self-reinforcing concentration. The city\u0026apos;s effective market access extends nationally through political and corporate networks. With 26% of China\u0026apos;s research personnel despite having 1.6% of population, Beijing demonstrates extreme spatial concentration of high-skill activities. Agglomeration economies manifest through thick labour markets with 750,000 university graduates annually, knowledge spillovers generating 45% of national patents in some technology classes, and specialised inputs including venture capital and professional services. Agglomeration also creates offsetting congestion costs including severe air pollution, traffic gridlock (average commutes of 52 minutes), and housing costs claiming 58% of average income.\u0026nbsp;\u003c/p\u003e"},{"header":"3. Methods","content":"\u003cp\u003e\u003cstrong\u003e3.1 Data Sources and Spatial Considerations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe use panel data from 1995-2015 compiled from Statistical Yearbooks of Lanzhou, Guiyang, and Beijing. This period captures China\u0026apos;s spatial transformation including migration flows of 150 million rural-urban migrants, transport infrastructure expansion that quadrupled the highway network, and evolving spatial policies. The timeframe provides sufficient variation to identify spatial mechanisms whilst maintaining institutional stability.\u003c/p\u003e\n\u003cp\u003eAll monetary variables are deflated using city-specific Consumer Price Indices, crucial given spatially heterogeneous inflation. Services in Beijing inflated 40% faster than goods in Lanzhou. Spatial price differences matter as identical nominal salaries represent vastly different real compensation across locations due to housing and service costs.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 Measuring Human Capital in Spatial Context\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe construct education and health indices using coefficient of variation weighting to reflect local conditions. For each indicator i in city j, we calculate:\u003c/p\u003e\n\u003cp\u003eEquation (3): CVij = \u0026sigma;ij / \u0026mu;ij\u003c/p\u003e\n\u003cp\u003eWhere \u0026sigma;ij is the standard deviation of indicator i in city j across the time period:\u0026mu;ij is the mean of that indicator over time. This temporal CV within each city reveals policy responsiveness given spatial constraints. An indicator stable in accessible Beijing but volatile in isolated Lanzhou suggests different development possibilities across space. The approach captures how spatial position shapes the ability to implement change.\u003c/p\u003e\n\u003cp\u003eEducation indicators span inputs including expenditure and salaries, processes including enrolment and infrastructure, and outputs including graduation rates, each affected differently by spatial factors. Distance increases costs through teacher recruitment difficulty whilst reducing benefits through graduate retention challenges. Health indicators similarly reflect spatial challenges as medical personnel concentrate in accessible cities whilst isolated areas struggle with basic provision.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 Spatial Econometric Considerations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur VECM specification incorporates spatial elements:\u003c/p\u003e\n\u003cp\u003eEquation (4): \u0026Delta;yjt = \u0026alpha;j + \u0026gamma;j(yj,t-1 - \u0026beta;\u0026apos;xj,t-1) + \u0026Sigma;i=1^p-1 \u0026Gamma;ji\u0026Delta;yj,t-i + \u0026delta;jMAjt + \u0026nu;jt\u003c/p\u003e\n\u003cp\u003ewhere \u0026Delta;yjt represents the change in variables for city j at time t, \u0026alpha;j captures city-specific constants, \u0026gamma;j is the error correction coefficient indicating adjustment speed to long-run equilibrium, \u0026beta; contains long-run relationship parameters, \u0026Gamma;ji captures short-run dynamics, MAjt represents market access from equation (2), and \u0026nu;jt is the error term. Market access for each city is calculated as the sum of distance-weighted GDP of all prefecture-level cities within 500km radius. All parameters (\u0026alpha;, \u0026gamma;, \u0026beta;, \u0026Gamma;, \u0026delta;) vary by city j, capturing how spatial position shapes economic relationships. While we cannot estimate spatial lag terms or test for spatial autocorrelation patterns with three cities, this specification allows spatial heterogeneity in all relationships.\u003c/p\u003e"},{"header":"4. Results","content":"\u003cp\u003e\u003cstrong\u003e4.1 Spatial Patterns in Human Capital Development\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFigure 1 illustrates the spatial configuration of our study cities.\u003c/p\u003e\n\u003cp\u003eLanzhou is located in northwest Gansu along the narrow Hexi Corridor, Guiyang in mountainous southwest Guizhou, and Beijing on the eastern coast, highlighting differing positions relative to major transport corridors and economic zones. Figure 2 reveals the spatial gradient in education resources, fundamental to understanding differential returns.\u003c/p\u003e\n\u003cp\u003eBeijing\u0026apos;s expenditure increased from approximately 12,000 to 50,000 CNY, whilst Lanzhou and Guiyang remain below 10,000 throughout the period.\u003c/p\u003e\n\u003cp\u003eThe tenfold expenditure gap reflects more than fiscal capacity. It represents the spatial concentration of quality education. Beijing attracts top teachers through 70% salary premiums, hosts elite schools, and benefits from corporate training programmes. This spatial inequality in education quality amplifies raw expenditure differences. Figure 3 illustrates demographic pressures varying across space.\u003c/p\u003e\n\u003cp\u003eThe figure shows Lanzhou and Guiyang with 800-1,100 students per 10,000 declining after 2007, whilst Beijing maintains 300-400 throughout. The spatial gradient in youth dependency reflects differential migration patterns. Beijing attracts working-age migrants whilst their children often remain in origin cities due to hukou restrictions, creating divergent demographic pressures. Interior cities must provide education for larger youth cohorts with fewer resources, representing a fundamental spatial inequality.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.2 Coefficient of Variation Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTable 1 reveals how different indicators drive change across spatial contexts.\u003c/p\u003e\n\u003ctable border=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"10\" style=\"width: 99px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable 1: Coefficient of Variation (CV) Results for Education Indicators\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIndicator\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 22px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLanzhou\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 22px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGuiyang\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 24px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBeijing\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eExpenditure per student\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e1,913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e2,226\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.86\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e2,086\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e2,420\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.86\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e10,241\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e27,896\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e0.37\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eTeacher salary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e16,875\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e22,199\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e17,607\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e24,025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e30,365\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e41,355\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e0.73\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003ePrimary students/10k\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e164\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e837\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e143\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e977\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e172\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e469\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e0.37\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eUniversity students/10k\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e465\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e690\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e211\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e494\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e0.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e442\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\n \u003cp\u003e693\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e0.64\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 28px;\"\u003e\n \u003cp\u003eSum of CV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e9.74\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 5px;\"\u003e\n \u003cp\u003e9.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 8px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 6px;\"\u003e\n \u003cp\u003e10.19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eSource: Authors\u0026apos; calculations. Full results for all 14 indicators available upon request.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe CV patterns reveal spatial constraints on development. Financial indicators show high variation in interior cities as governments struggle to increase funding from low bases. Beijing\u0026apos;s lower expenditure CV of 0.37 reflects stable high investment. Demographic indicators vary more in Beijing due to migration volatility, revealing how accessibility creates different development challenges.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.3 Human Capital Returns Across Space\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFigure 4 demonstrates the remarkable convergence in education indices despite persistent resource gaps.\u003c/p\u003e\n\u003cp\u003eThe figure shows indices converging from initial gaps where Beijing starts at 47, Lanzhou at 29, and Guiyang at 27, to a narrow final range of 65-73.\u003c/p\u003e\n\u003cp\u003eConvergence masks fundamental spatial differences in what drives improvement. Interior cities achieved gains through basic quantity expansion including more schools, teachers, and enrolment. Beijing improved through quality enhancement including better teacher training, curriculum innovation, and technology integration. The indices converge whilst the nature of education diverges spatially.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4 Econometric Evidence of Spatial Heterogeneity\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTable 2 presents our core findings on spatially differentiated returns.\u003c/p\u003e\n\u003ctable border=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" style=\"width: 99px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable 2: Long-Run VECM Coefficients Revealing Spatial Heterogeneity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLanzhou\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGuiyang\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBeijing\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003eln(Population)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e-8.33***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e6.20***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e1.02***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e(2.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e(2.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e(0.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003eln(Education)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e1.23***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e0.93***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.61**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e(0.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e(0.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e(0.13)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003eln(Health)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.27**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e-0.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e-0.10\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e(0.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e(0.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e(0.15)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003eln(Fixed Assets)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.74***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e0.30***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.77***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e(0.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e(0.07)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e(0.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003eln(R\u0026amp;D)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e0.22***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.47***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 36px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 21px;\"\u003e\n \u003cp\u003e(0.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 20px;\"\u003e\n \u003cp\u003e(0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e(0.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003e*Note: Standard errors in parentheses. **, ** denote 1%, 5% significance. Source: Authors\u0026apos; VECM estimations.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe results reveal spatial heterogeneity. The gradient from 1.23 in isolated Lanzhou to 0.61 in accessible Beijing demonstrates how distance amplifies returns. In remote locations, educated workers are scarce, creating high marginal productivity. The 100% differential suggests spatial inefficiency where the same education investment generates double the growth impact in lagging regions.\u003c/p\u003e\n\u003cp\u003ePopulation coefficients from -8.3 to +6.2 reveal how space shapes migration-development relationships. Lanzhou\u0026apos;s negative coefficient indicates vicious cycles where education enables out-migration, undermining local development. Guiyang\u0026apos;s positive coefficient suggests it has reached spatial thresholds where improved human capital attracts rather than repels talent. Beijing\u0026apos;s modest effect reflects spatial equilibrium where in-migration benefits balance congestion costs. R\u0026amp;D returns increase monotonically with spatial position from 0.02 to 0.47, confirming agglomeration theories. Innovation requires dense networks of researchers, specialised suppliers, and knowledge institutions that concentrate spatially. Isolated cities cannot generate returns from R\u0026amp;D without the spatial preconditions.\u003c/p\u003e"},{"header":"5. Discussion","content":"\u003cp\u003e\u003cstrong\u003e5.1 Theoretical Implications for Spatial Economics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur findings extend spatial economic theory. The 100% education return differential between remote and accessible cities quantifies distance penalties beyond transport costs. This gradient reflects how distance limits idea flows, reduces matching quality in labour markets, and constrains specialisation possibilities. Standard spatial models focusing on goods transport miss these knowledge-distance interactions crucial for understanding modern spatial inequality.\u003c/p\u003e\n\u003cp\u003eBeijing's lower education returns appearparadoxical through an agglomeration lens. However, this reflects spatial equilibrium where agglomeration benefits capitalise into land prices and congestion costs, reducing net returns. The spatial sorting of high-ability workers to Beijing means marginal education investments have less impact than in human capital-scarce locations. Health returns appearing only in high-mortality locations reveal spatial development thresholds. Basic infrastructure must reach minimum levels before locations can exploit agglomeration advantages. This sequencing of first overcoming fundamental disadvantages, then building specialisation, challenges models assuming smooth spatial transitions.\u003c/p\u003e\n\u003cp\u003eThe interaction between education and migration in remote cities creates spatial poverty traps missed by standard models. Education investments intended to develop lagging regions may accelerate divergence by facilitating out-migration. This paradox requires reconceptualising spatial policy. Isolated investments fail without complementary improvements in spatial connectivity and local opportunity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.2 Policy Implications\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur baseline evidence provides quantitative guidance for spatially differentiated policies. The 100% return differential strongly supports Common Prosperity's emphasis on spatial redistribution. Each CNY invested in education in cities like Lanzhou generates roughly double the return compared to Beijing. However, population results warn against isolated interventions. Successful spatial policy targets education, employment, and connectivity simultaneously. Supporting literature confirms these patterns. An (2018) finds that rural teachers earn wages 7-41% lower than the provincial average, with one study showing highest teacher wages in Shanghai at RMB 62,300 per year compared to Henan Province at RMB 12,800 per year. World Bank (2022) analysis of Chinese education spending shows coastal provinces spend 3-5 times more per student than western provinces. Our findings suggest spatially differentiated investment sequences. Remote cities like Lanzhou require priority on reducing distance penalties through e.g. digital infrastructure and transport links before major human capital investments. Transitional cities like Guiyang benefit from balanced packages combining moderate education and R\u0026amp;D with connectivity improvements. Agglomerated cities like Beijing can focus on managing congestion whilst maintaining innovation advantages.\u003c/p\u003e\n\u003cp\u003ePost-2015 digital transformation may fundamentally alter spatial relationships. Remote work capabilities could reduce distance penalties for knowledge workers. E-commerce platforms may overcome market access constraints for isolated regions. Digital education could narrow quality gaps between locations. However, agglomeration advantages in innovation and high-skill services may intensify, creating new forms of spatial inequality. Future research should examine whether digital transformation reduces or reshapes spatial heterogeneity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.3 Methodological Contributions to Spatial Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe representative city approach offers advantages. By selecting cities representing broader spatial categories, we offer analytical depth whilst maintaining external validity. The 247 cities sharing our representatives' characteristics can apply insights from this analysis.\u003c/p\u003e\n\u003cp\u003eLarge-N spatial econometrics identify correlations yet struggle with mechanisms. Our representative-selection approach reveals how distance creates teacher recruitment challenges, why health thresholds matter spatially, and when migration reinforces versus undermines development. Researchers may identify 3-5 cities representing key spatial positions including remote resource regions, intermediate manufacturing, and connected services, then trace mechanisms through more detailed comparison.\u0026nbsp;\u003c/p\u003e"},{"header":"6. Conclusions","content":"\u003cp\u003eThis paper reveals spatial heterogeneity in human capital returns across China\u0026apos;s urban hierarchy. Examining three representative cities (representative of \u0026nbsp;247 urban economies occupying different spatial positions), we quantify how distance, density, and connectivity shape development possibilities. Education returns varying from 1.23 in remote Lanzhou to 0.61 in accessible Beijing demonstrate 100% spatial differentials that challenge uniform national policies. These patterns reflect not simply different development levels but fundamental spatial mechanisms including distance penalties limiting idea flows, agglomeration advantages creating congestion, and migration dynamics that can either reinforce or undermine local development. Table 3 synthesises our key findings, connecting them to existing literature and policy.\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable 3: Summary of Findings, Literature Links, and Policy Implications\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eFinding\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eOur Results\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eSupporting Literature\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003ePolicy Connection\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eEducation return differentials\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e100% higher returns in Lanzhou (1.23) vs Beijing (0.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eFleisher et al. (2010): 60% provincial gaps; Gennaioli et al. (2013): 0.3-1.8 range internationally.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSupports targeted western investment; validates 3-5x spending gaps identified by World Bank (2022).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eHealth threshold effects\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eReturns only where mortality \u0026gt;6/1,000 (Lanzhou: 0.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eBloom et al. (2004): basic health as growth prerequisite; Shen, Q., Chang et al., (2020) similar mortality thresholds.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eHealthy China 2030: Focus basic health in high-mortality areas before quality improvements.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003ePopulation-migration traps\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e-8.33 in Lanzhou, +6.20 in Guiyang, +1.02 in Beijing.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTombe \u0026amp; Zhu (2019): hukou creates spatial misallocation; Au \u0026amp; Henderson (2006): migration restrictions reduce efficiency.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eHukou reform priorities: Bundle education with local opportunities to prevent brain drain.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eDistance-R\u0026amp;D interaction\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eZero returns in Lanzhou, 0.47 in Beijing.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eMoretti (2011): innovation requires density; Krugman (1991): agglomeration necessary for knowledge spillovers.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eInfrastructure must precede innovation investment in corridor cities.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eSpatial convergence patterns\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eEducation indices converge while quality diverges.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLi \u0026amp; Huang (2009): provincial convergence; our city-level analysis reveals quality divergence.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eWestern Development Strategy: Quantity targets achieved, shift focus to quality enhancement.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eSource: Authors\u0026apos; synthesis of empirical findings with literature and policy documents.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eOur representative city analysis provides spatial insights that complement large-N spatial econometric studies, demonstrating the value of mixed methodological approaches in spatial economics. While large samples identify average effects, representative analysis points to spatial mechanisms and thresholds crucial for policy design. The coefficient of variation approach captures how spatial position shapes policy responsiveness, whilst our econometric strategy allows all parameters to vary spatially.\u003c/p\u003e\n\u003cp\u003eOur three-city analysis cannot model explicit spatial spillovers or test for spatial autocorrelation patterns that larger samples would permit. This limitation means our findings represent location-specific relationships rather than spatially integrated effects. Future research should employ explicit spatial econometric models with spatial weight matrices to capture network dependencies our approach cannot address.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eSpecifically, future research should employ spatial panel models with contiguity-based weight matrices to capture inter-city spillovers and test for spatial autocorrelation in human capital returns.\u003c/p\u003e\n\u003cp\u003eThe findings establish quantitative benchmarks for evaluating spatial policies under Common Prosperity and Belt and Road initiatives. Substantial return differentials support spatial targeting, but population dynamics warn against isolated interventions.. The threshold nature of returns suggests spatial sequencing where regions must first overcome basic disadvantages, then build human capital, and finally develop innovation capacity.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePost-2015 digital transformation including remote work, e-commerce, and online education may reshape distance penalties and agglomeration advantages. Whether these impact \u0026nbsp;spatial inequality or create new forms of concentration remains an empirical question requiring updated analysis. Our methodology provides a replicable framework for analysing spatial heterogeneity in other developing countries facing regional inequality. Researchers can identify 3-5 cities representing key spatial positions, apply coefficient of variation weighting to capture local dynamics, and use VECM to identify spatially varying relationships. This approach is particularly valuable for countries with diverse geography where aggregate analysis masks crucial spatial variation.\u003c/p\u003e\n\u003cp\u003eAs China and other nations pursue spatially balanced development, recognising and responding to spatial heterogeneity rather than implementing spatially blind policies offers a promising path toward inclusive growth.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAn, X. H. (2018). Teacher salaries and the shortage of high-quality teachers in China\u0026apos;s rural primary and secondary schools. Chinese Education \u0026amp; Society, 51(2), 103-116. https://doi.org/10.1080/10611932.2018.1433411\u003c/li\u003e\n\u003cli\u003eAu, C. C., \u0026amp; Henderson, J. V. (2006). Are Chinese cities too small? Review of Economic Studies, 73(3), 549-576. https://doi.org/10.1111/j.1467-937X.2006.00387.x\u003c/li\u003e\n\u003cli\u003eBehrens, K., Duranton, G., \u0026amp; Robert-Nicoud, F. (2014). Productive cities: Sorting, selection, and agglomeration. Journal of Political Economy, 122(3), 507-553. https://doi.org/10.1086/675534\u003c/li\u003e\n\u003cli\u003eBloom, D. E., Canning, D., \u0026amp; Sevilla, J. (2004). The effect of health on economic growth: A production function approach. World Development, 32(1), 1-13. https://doi.org/10.1016/j.worlddev.2003.07.002\u003c/li\u003e\n\u003cli\u003eCombes, P. P., D\u0026eacute;murger, S., \u0026amp; Li, S. (2019). Migration externalities in Chinese cities. European Economic Review, 76, 152-167. https://doi.org/10.1016/j.euroecorev.2015.02.004\u003c/li\u003e\n\u003cli\u003eFleisher, B., Li, H., \u0026amp; Zhao, M. Q. (2010). Human capital, economic growth, and regional inequality in China. Journal of Development Economics, 92(2), 215-231. https://doi.org/10.1016/j.jdeveco.2009.01.010\u003c/li\u003e\n\u003cli\u003eGennaioli, N., La Porta, R., Lopez-de-Silanes, F., \u0026amp; Shleifer, A. (2013). Human capital and regional development. Quarterly Journal of Economics, 128(1), 105-164. https://doi.org/10.1093/qje/qjs050\u003c/li\u003e\n\u003cli\u003eGlaeser, E. L., \u0026amp; Lu, M. (2018). Human capital externalities in China. NBER Working Paper No. 24925. https://doi.org/10.3386/w24925\u003c/li\u003e\n\u003cli\u003eGlaeser, E. L., \u0026amp; Resseger, M. G. (2010). The complementarity between cities and skills. Journal of Regional Science, 50(1), 221-244. https://doi.org/10.1111/j.1467-9787.2009.00635.x\u003c/li\u003e\n\u003cli\u003eKrugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99(3), 483-499. https://doi.org/10.1086/261763\u003c/li\u003e\n\u003cli\u003eLi, H., \u0026amp; Huang, L. (2009). Health, education, and economic growth in China: Empirical findings and implications. China Economic Review, 20(3), 374-387. https://doi.org/10.1016/j.chieco.2008.05.001\u003c/li\u003e\n\u003cli\u003eLiu, J., Wang, M., Yang, L., Rahman, S., \u0026amp; Sriboonchitta, S. (2024). Multidimensional spatial inequality in China and its relationship with economic growth. Humanities and Social Sciences Communications, 11(1), 1-15. https://doi.org/10.1057/s41599-024-03961-y\u003c/li\u003e\n\u003cli\u003eMarshall, A. (1890). Principles of Economics. London: Macmillan.\u003c/li\u003e\n\u003cli\u003eMoretti, E. (2011). Local labor markets. In Handbook of Labor Economics (Vol. 4, pp. 1237-1313). Elsevier. https://doi.org/10.1016/S0169-7218(11)02412-9\u003c/li\u003e\n\u003cli\u003eRedding, S., \u0026amp; Venables, A. J. (2004). Economic geography and international inequality. Journal of International Economics, 62(1), 53-82. https://doi.org/10.1016/j.jinteco.2003.07.001\u003c/li\u003e\n\u003cli\u003eShen, Q., Chang, B., Yin, G., \u0026amp; Wang W. (2020). The Impact of Health Investment on Economic Growth: Evidence from China. \u003cem\u003eIranian Journal of Public Health\u003c/em\u003e, \u003cem\u003e49\u003c/em\u003e(4), 684\u0026ndash;692.\u003c/li\u003e\n\u003cli\u003eTombe, T., \u0026amp; Zhu, X. (2019). Trade, migration, and productivity: A quantitative analysis of China. American Economic Review, 109(5), 1843-1872. https://doi.org/10.1257/aer.20150811\u003c/li\u003e\n\u003cli\u003eWei, Y. D. (2000). Regional Development in China: States, Globalization and Inequality. Routledge. \u003c/li\u003e\n\u003cli\u003eWorld Bank (2019). China Transport Sector Assessment. Washington, DC: World Bank.\u003c/li\u003e\n\u003cli\u003eWorld Bank (2022). China - Gansu Technical and Vocational Education and Training Project. Washington, DC: World Bank.\u003c/li\u003e\n\u003cli\u003eZhu, J., Zhu, M., \u0026amp; Xiao, Y. (2019). Urbanization for rural development: Spatial paradigm shifts toward inclusive urban-rural integrated development in China. \u003cem\u003eJournal of Rural Studies\u003c/em\u003e, \u003cem\u003e71\u003c/em\u003e, 94\u0026ndash;103. https://doi.org/10.1016/j.jrurstud.2019.08.009\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Spatial heterogeneity, Human capital returns, Distance penalties, Regional convergence, China, Spatial econometrics, VECM","lastPublishedDoi":"10.21203/rs.3.rs-7216194/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7216194/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eChina's coastal cities enjoy per capita GDP ten times higher than western interior cities despite decades of spatial integration policies. We examine three representative cities, Lanzhou, Guiyang, and Beijing, exemplifying 127 resource-dependent, 89 transitioning, and 31 advanced urban economies respectively from 1995-2015. Using coefficient of variation weighting and Vector Error Correction Models, we identify education returns of 1.23 in resource-dependent Lanzhou versus 0.61 in advanced Beijing, demonstrating 100% spatial differentials. Health investments show threshold effects only where mortality exceeds 6 per 1,000. Population effects vary from -8.33 in isolated western cities to +6.20 in accessible transitional economies. While our three-city analysis cannot capture spatial spillovers between cities, city-specific parameters reveal how location fundamentally shapes development relationships. These baseline findings provide quantitative benchmarks for evaluating contemporary spatial policies under China's Common Prosperity initiative.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eJEL Classification:\u003c/strong\u003eR11, C31, O15, R23\u003c/p\u003e","manuscriptTitle":"Distance Penalties and Human Capital Returns: A Spatial Analysis of Chinese Cities","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-15 08:28:10","doi":"10.21203/rs.3.rs-7216194/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c35f518d-6410-4c7f-ba74-c0ada5cec129","owner":[],"postedDate":"December 15th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-15T08:28:10+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-15 08:28:10","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7216194","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7216194","identity":"rs-7216194","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}